What is MAI-Image-2.5?

MAI-Image-2.5 is a text-to-image and image editing model that handles localized edits, identity preservation, and text rendering. It ranks No. 2 on Arena’s Image Edit leaderboard, ahead of Nano Banana 2.1, and is available via Foundry and OpenRouter for developers building production image workflows. The model also powers Microsoft products like PowerPoint and OneDrive, enabling high-quality generation and precise, controllable editing at scale.

Who it's for

• Developers building production image workflows that need reliable text-to-image generation and fine-grained editing capabilities.
• Product teams integrating image editing into apps or services, requiring identity preservation and localized edits without breaking the rest of the image.
• Enterprise users leveraging Microsoft tools like PowerPoint and OneDrive for presentation-ready visuals and precise photo editing.

Key features

Step-change in text-to-image quality

MAI-Image-2.5 produces more detailed, coherent images from prompts, with stronger text rendering, product imagery, and prompt adherence. It delivers a +75 point improvement over MAI-Image-2 on Arena scores.

Complex visual reasoning

The model understands scene structure, lighting, scale, and spatial relationships. This allows it to make edits that fit the image context, such as adding an object with the right perspective and shadows.

Fine-grained edit control

MAI-Image-2.5 supports precise, localized edits — from replacing an object or updating text to removing motion blur — without changing the rest of the image. This makes it suitable for production editing workflows.

Face and identity consistency

The model preserves facial identity across edits, maintaining recognizable likeness even through changes in pose, expression, or viewpoint. This is critical for applications requiring consistent character or person imagery.

What stands out

MAI-Image-2.5 ranks No. 2 on Arena’s Image Edit leaderboard, ahead of Nano Banana 2.1, and delivers a +75 point improvement over its predecessor.

This ranking reflects strong performance across 12 editing categories, with the largest gains in Text Rendering (+107) and Cartoon, Anime & Fantasy (+90). The model also powers Microsoft products like PowerPoint and OneDrive, making it one of the few image models integrated directly into enterprise productivity tools.

Worth checking out if…

You need a production-ready image model that combines high-quality generation with precise, localized editing. It’s especially relevant if you’re building workflows that require identity preservation, text rendering, or complex visual reasoning — or if you want to leverage the same model powering Microsoft’s PowerPoint and OneDrive features.